[zt]PowerBuilder History - A Canadian Perspective With An Aussi Twist
PowerBuilder History - A Canadian Perspective With An Aussi Twist
When did Sybase develop the product and how did it evolve?
By: Chris Pollach
Jan. 8, 2005 03:45 PM
I have been asked many times by various clients, students, and the IT curious about PowerBuilder: When did Sybase develop the product and how did it evolve? I keep telling this story and answering e-mails on the subject. I am now to the point where I have decided that I should have PBDJ formally publish this story for posterity. This story is solely from my own personal perspective and I'm sure that I may have a somewhat distorted view of time and space (which keeps getting worse at my age). But anyway, here is a perspective of how life (as a DataWindow) began.
Episode I - The Phantom Product
In a galaxy, far, far away... before Sybase... PowerBuilder was developed as a prototype by Cullinet Database Systems of Boston. Cullinet was the developer of the IDMS DBMS and ADS-Online (Application Development System). IDMS was originally developed by BF Goodrich, the tire company, in London, Ontario, Canada, and sold to Cullinane (that later changed their name to Cullinet) located in Boston, Massachusetts. They had an early PC product called "Golden Gate," which proved that you could do GUI client/server application development with mainframe DBMSs. Originally DOS based, Golden Gate was then converted to run under MS Windows. Cullinet also realized that PC-based DBMSs and development tools were on the immediate horizon (1984). Cullinet had an enormous success with ADS-Online (327x-based RAD development tool) and wanted to see if a similar GUI-based tool could be developed. The main features of ADSO included RAD; real-time design, programming, compiling, and debugging; and interactive prototyping. It also used a centralized Data Dictionary, interfaced with various CASE tools, and could deploy to production from development.
In 1984, when I was the technical support manager for Cullinet Canada, Cullinet started their Personal Computer ADSO prototype, which would later become PowerBuilder. The project leader was Dave Litwack, who was in charge of the ADSO product and IDMS-DC (data communications - teleprocessing system and CICS equivalent product). Dave had a great understanding of RAD development tools and telecommunications because of his Cullinet experience. The new product was to have the same key functionality as ADSO (interesting that ORCA was basically working in 1985 in the PowerBuilder prototype because ADSO/IDMS had it), but also added a real key feature: "a smart data aware object." At that time Cullinet was experimenting with a feature called LRF (Logical Record Facility) and DB stored procedures. This object would encapsulate data handling away from the application, but would be a client piece, so there was no dependency on any DBMS.
For the first part of the prototype, Dave chose a real keen "C" programmer named Kim Sheffield. John Griffin - a friend of mine from Ottawa, Ontario - was also recruited by Dave. John was an excellent mainframe assembler programmer and wanted to cross over to C. Dave had him build the "Menu" painter. In later years, John married Julie, another Cullinet developer, and she would help rewrite the Menu Painter and add Remote Debugging to PB for EAServer.
Dave also wanted to have the new tool that was fully OO - object oriented. The C++ language was coming on strong and SmallTalk was the "talk of the town" for serious OO programmers. Dave wanted PowerBuilder to adopt the SmallTalk OO principles but make it easy for the business developer to use.
In 1985 a crude prototype was shown to the Cullinet inside circle. The product's potential was immediately expounded by senior management (one of those being Bobby Orr - the hockey legend [another good story I can tell you sometime] - who was on the board of directors of Cullinet at that time). Unfortunately, Cullinet had serious challenges for various takeover bids by different companies, including CA (Computer Associates). CA at that time had already purchased DataCom and wanted Cullinet for IDMS. CA's mentality then was to buy out the competition, sell off any non-key products, and milk the maintenance contracts of key clients. With little or no development personnel (fired), there was no overhead and all profit. In 1986, CA was successful on a hostile takeover of Cullinet. The new PC product was considered non-essential and all developers were let go (that is why today I will never buy a CA product)!
问题点数：0、回复次数：3Top
1 楼free_soft（）回复于 2006-03-03 21:52:10 得分 0
Episode II - The Little Droid that Could
In 1986, PowerSoft was developing business applications for the VAX platform. PowerSoft also realized the PC development arena was about to explode and started looking around for a leading-edge GUI development tool. They hired an independent consultant, Dave Litwack, to help advise them on exactly what they should be looking for. At that time, Gupta's SQLWindows was the only serious product. Other than that you had to get down and code "C," which was not what PowerSoft wanted business programmers to use. Dave mentioned his involvement with Cullinet and their last prototyping effort. PowerSoft approached CA and asked if they could acquire the prototype code (originally done in C). CA said that they had looked at the PB prototype and that they concluded that there was no future in it (duh) - so give us a few bucks for the code and good luck with it!
In 1988, three years after the original concept prototype, PowerSoft had the code and Dave (now hired by PowerSoft) was able to hire the other programmers who worked on the original prototype (what a fluke) as they were looking for some challenging work at that time as well. PowerSoft then christened the new product "PowerBuilder" and began to enhance the code. Since they were a business solutions developer, they used PowerBuilder internally to recode and replace their VAX products. Testing was "hands-on" and very intensive under real-world developer scenarios. To get funding for this intense effort, PowerSoft partnered with HP. HP gave them a blank check after seeing a demonstration (they are also responsible for the Tilde "O" format [~Onn] for "octal" as HP was an 8-bit machine in those days). PowerBuilder became an internal standard at HP.
At Microsoft's Redmond office, the people in charge of internal systems were faced with the same problems PowerSoft was trying to resolve - they needed a tool for business developers. They contacted their friends at HP and were told that PowerBuilder was the only up-and-coming tool they should look at. In early 1989, Microsoft purchased licenses for PB and was the second worldwide user. The "Royal Australian Air Force" was the first official user - makes me proud as I'm an Australian from Cooma, NSW.
When I was four years old my dad took me to the University of Sydney where he was using the SILLIAC I (Sydney version of the Illinois Automatic Computer) - the first commercial computer ever built (http://members.iinet.net.au/~dgreen/silliac.html) to do the stress and strain calculations on the Snowy Mountains Hydro Electric power dams (largest hydro generation in the world even today: www.snowyhydro.com.au ), which is where I met the first "debugger" in the world ... but that's another story. The Snowy Hydro is still an active PowerBuilder site even today. Even Microsoft used PowerBuilder for their Inventory Management System, MS University Scheduling System, etc., and was "blown" away with its data prowess - especially this new object called the "DataWindow" (thanks Kim).
Top
2 楼free_soft（）回复于 2006-03-03 21:53:36 得分 0
Episode III - The "Force"
In 1989 I was doing a project for the Canadian government on behalf of Revenue Canada and Treasury Board. My task was to evaluate emerging RDBMS technology and propose the top three to be recommended to all government departments. I completed that in the late summer of 1989. One of the contenders was Microsoft's SQLServer (which was a port of the release 4.x version of Sybase's SQLServer to the OS/2 platform). To verify the final benchmark results, I had a representative from each DBMS vendor drop in and tweak their environments and concur with my approach and results.
To that end, a Microsoft engineer from New Hampshire came to Ottawa. He was extremely helpful and asked about what I was going to do next. I told him that the next phase was to review and recommend GUI development tools for the three top recommended RDBMS products. This is when the guy floored me; he said that we should call this new company called PowerSoft and get PowerBuilder because all his developer buddies at Redmond were using PowerBuilder (certainly not what the MS salesman was saying). I called PS, but they said I could not get an evaluation copy, I had to buy it - but I could return it if I didn't like it (if any of you have worked for the U.S. or Canadian governments, you know what kind of stupid remark that would be). But they told me the history of PB so far and about hiring an X-development team from a DBMS company in Boston who was bought out by CA. I thanked the salesperson, hung up, hit the redial button, and asked for Dave Litwack, where I was immediately put through - to my pleasant surprise. Dave said, "qaStaH nuq? Chris!" (Klingon for "What's happening Chris?" www.kli.org/tlh/phrases.html) as the response and the next day a copy of PB was on my desk (thanks again Dave).
I passed the release (version 0.8 Alpha, which came on two diskettes at the time) to my co-developers and they pounded at it for three days and took it for a spin with the top RDBMSs we had in place. PB's DataWindow was like a "breath of fresh air" when it came to data handling and SQL generation compared to any tool we had touched thus far. The speed was also very close to C and made SQLWindows look like molasses in January (Canadian perspective - eh).
PB became the premier product of the top three development tools recommended to the Canadian Government. Revenue Canada used the product to build the GST (Government Sales Tax) processing system, which captures and tracks all GST Tax returns even to this day - talk about mission critical. Other departments soon dove in and, today, most Canadian Government departments use PB for their mission-critical systems: coming into Canada your license plate is scanned on your car (checked with a DB managed by PB) and/or your passport is scanned (all done with PB); if your plane lands on a Canadian runway (billed by PB working in concert with the radar system - 24x7 operations), if you have an Old Age pension (front end all done in PB - released January 2002), log a case with the Supreme Court or use their Web site: http://www.scc-csc.gc.ca, open a case with the Tax Court of Canada, register a firearm, deploy UN troops, get a security clearance with DND or PWGSC, participate in a Federal Election, and many, many more!
In the early 1990s SQLWindows and a new product - Delphi - took a run at PB from the performance side. SQLWindows added C generation and Delphi came like that out of the box. This made SQLWindows as fast as PB, but Delphi blew PB away in looping operations. The DataWindow - 90% C and 10% assembler - still blew the other products away (I have a great story how last year on VB.net versus PB 9 at a Government department using MS-SQLServer, PB blew VB away by 4000% better performance - but getting back to our history lesson...).
Dave, Bill Rabkin ("the" original PB technical evangelist), and I had discussions about compilers for many years. We often remarked how efficient the Waterloo, Ontario, guys were at building top-of-the line compiler technology. At the same time Gupta (SQLWindows) started shipping its own RDBMS (SQLBase) with its own product. PowerSoft wanted to do the same to match the competition. It came as no surprise to me when they started shipping the WatCom SQL RDBMS with PB (I think that was release 2). Like Victor Kiam ("I liked it so much I bought the company"), that is exactly what PowerSoft did for release 3. But, a hidden gem appeared to PB right after that, when Dave said to the C compiler guys at WatCom, "Could you take PB's P-Code and generate pure C (PB at that time had moved from C to C++). The WatCom people said: "Sure," and had it working within a week. This really made PB "toast" Delphi in performance and with the DataWindow - leaving them in the dust (and it still does today).
At the same time, Bill Gates came to Ottawa to deliver a keynote address to the Canadian Government. I met Bill and he informed me (in 1994) of some interesting facts: MS uses the WatCom C compiler for the VB JET engine, some of MS-Access, and all of FoxPro for Windows. MS could not convert the products to use their own C compiler as it was 400 time slower than WatCom's and the user community would not stand for the performance loss. A friend of mine who was hired out of Toronto to work in Redmond told me that Bill wanted VB to be fully OO and he had a prototype (1993-4), but upon demonstrating the product, key business users would not accept the necessity to completely rewrite the code (like VB 6 to 7 [.NET] programmers have now) in order to properly derive full OO benefits. They told MS that they were better off with PowerBuilder. Recently, many VB 7 programmers here in Ottawa have told me that they're recommending that their departments go to PowerBuilder as it is much more OO friendly and the learning curve is substantially lower (interesting comments?).Top
3 楼free_soft（）回复于 2006-03-03 22:03:36 得分 0
Episode IV - The Attack of the Corporate Clones
In the mid 1990s Gupta was crushed by Oracle's multiple attempts to do a hostile takeover. Oracle wanted to compete against PB, but PowerSoft was untouchable at that time due to its financial stability. So they went after SQLWindows to replace their development suite (SQL Forms, etc.). Even today, any student of mine who has developed in SQL Forms and then sees PB, he or she drops it like a hot potato. Here's a recent example: I worked on a new system that was developed by two Oracle developers on an Oracle DB. They worked for nine months trying to build this new system and could not even get a prototype done. I worked on the system for three months with PB 7/8 and had the full system working within three months. This blew away the Oracle developers even on their own DBMS! The application is currently running under PB 10.2 after being ported a few months ago.
Dave Litwack and the PowerSoft executives were very nervous about Oracle's actions (shades of CA déjà vu), and wanted to band together with a larger company to make sure another CA would not happen to them. Sybase had helped port the SQLServer DBMS over to the MS-NT Platform and knew of PB's prowess (even today 63% of all Oracle DBMS sites use PB as the development tool), but lacked any good GUI development tool. The two companies merged (and brought WatCom along) to better complement their technologies into one company offering. The WatCom Company was hence renamed to iAnyWhere Solutions.
Dave Litwack and Kim Sheffield left Sybase shortly after that to develop a really "cool" product known as SiverStream (http://jdj.sys-con.com/read/36628.htm). Again, with the key concepts of an open IDE, integrated tools, work with any DBMS, central Data Dictionary, ease of deployment to production, RAD prototyping, service objects, etc. Sybase would be wise to not lose sight of these key aspects of a good IDE. In recent days, Microsoft has finally learned this expensive lesson (it took them over a decade) with Visual Studio 2005. I used the product and became certified for this new Java tool in late 1998-99. A few Canadian Federal Government agencies started using it and we also developed a nice Web portal in SilverStream. One of the key features that Kim added to SS was a DataWindow-like object that supported a TreeView look and feel (PB 10.5 eat your heart out - this was running in 1999-2000). The components could be used for a native Java application or through a Web browser. All these features placed SS far ahead of PowerBuilder and at that time PowerJ. Looking at this in hindsight, it's too bad John Chen let these people go instead of continuing their work on PB and now the new "WorkSpace" product. Sybase lost important momentum in the PB area by also focusing only on Java in the early 2000 years.
In 2003, SilverStream was bought by Novell where Dave Litwack still resides. Kim Sheffield has left Novell and is now the principle owner of fyiReporting Software LLC. (www.fyireporting.com/company.html). He has an interesting reporting product written in C# that could be plugged into a PB application (what goes around...). Since I really respect Kim as a developer, it's interesting to see that he has abandoned Java for the .NET world. Maybe Sybase should keep an eye on key developers like Kim as they reflect current trends in the market place (my $0.02 worth). Dave is now a senior VP and general manager at Novell in charge of the Identity Driven Products Group (www.novell.com/company/bios/litwack.html). Bill Rabkin, who left Sybase for Rational, but no longer works there after the IBM merge, is now a WorldServer product evangelist with Idiom Technologies.
Episode V - PowerBuilder Strikes Back
During the PowerSoft/Sybase merge, however, the "tools innovation" direction of PowerSoft was lost (I believe due to the "Server" product mentality at the senior management level). But recently, we can see the recommitment to PowerBuilder in the form of PB 10, 11, etc., the long-term "blueprint" for PB 12.0, the downturn of Java (new statistics for the last two years stated that 60% of all U.S. companies that started a Java project last year canned it), introduction of the PocketBuilder product to address the Windows CE (Pocket PC) direction, etc.
Interestingly enough the PocketBuilder product development is headed up by Reed Shilts (a long time PowerBuilder guru) and John Griffin - from the original Cullinet days and a native Ottawan - still adding some great Canadian content to the PB product line. The iAnyWhere division of Sybase (old WATCOM) subsidiary here in Waterloo, Canada, is turning out some great new products lately and is still producing what I refer to as the best small to medium-sized DBMS available in the market today. Maybe this is a real key turning point to the development refocusing efforts that I have seen lately at Sybase (but that's another story).
Top
